Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Os Rash
1. Iptables Attack Visualization
Michael Rash
Security Architect
Enterasys Networks, Inc.
http://www.cipherdyne.org/
OSCON
20070726
Copyright (C) 2007 Michael Rash 1
2. Agenda
Why visualize iptables log data?
●
context and change
–
iptables log message formats and data
–
completeness
New psad2.0.8 release with Gnuplot interface
●
Honeynet Project Scan34 iptables
●
visualizations (see http://www.honeynet.org)
Gnuplot
–
AfterGlow
–
Copyright (C) 2007 Michael Rash 2
3. Why Visualize Data?
Consider the following set of values (each value
●
represents the number of packets to unique
TCP ports per minute):
5, 4, 2, 1, 3, 4, 55, 58, 70,
85, 120, 9, 2, 3, 1, 5, 4
Copyright (C) 2007 Michael Rash 3
5. Graph Types
Gnuplot
●
Dots, points, lines through points – Good at seeing
–
trends and relationships in large data sets (>
100,000 points)
3D mode with interactive viewing
–
AfterGlow
●
Link Graphs – Good at expressing network
–
relationships with IP addresses
Copyright (C) 2007 Michael Rash 5
6. Iptables Log Data
Graphs will only be as good as the data source
●
permits
Iptables logs include nearly every interesting
●
field for both the network and transport layer
headers – we just need an effective way to
parse and then visualize this data
Iptables log messages can be tied to
●
application layer string matches via the string
match extension (see fwsnort)
Copyright (C) 2007 Michael Rash 6
14. psad2.0.8 release
Interfaces with Gnuplot – parses iptables log data
●
according to header field requirements and builds both a
data file and a directives file for Gnuplot
Various counting modes are supported across different
●
time scales
Granular requirements on iptables field values (including
●
negation)
Gnuplot works best with integer data (separated by
●
commas or spaces), so IP addresses need to be
translated into integer equivalents
Copyright (C) 2007 Michael Rash 14
15. psad2.0.8 release (cont'd)
psad gnuplot command line arguments:
●
gnuplotgraphstyle “points”, “dots”, “linespoints”
gnuplotfileprefix <file>.gnu, <file>.dat, <file>.png
gnuplottitle – Graph title (used for psad command line)
gnuplotsortstyle – time sorting vs. value sorting
gnuplot3D – Use “splot” for threedimensional viewing
gnuplotview – Set viewing angle
Copyright (C) 2007 Michael Rash 15
16. The Honeynet Scan34 Data Set
The goal of a Honeynet is to collect data on real
●
compromises – packet traces and log data is an excellent
learning tool
Over 170,000 lines of iptables log data in the Scan34
●
challenge – covers a five week period
Contains evidence of port scans, port sweeps, worm traffic
●
(Slammer, Nachi), and an outright compromise of a
system – we will visualize these graphically
All IP addresses are sanitized to 11.11.0.0/16
●
http://www.honeynet.org/
●
Copyright (C) 2007 Michael Rash 16
17. Port Scans
Packets to many different ports from a single
●
source address
Not just SYN packets, think of Nmap sF, sN,
●
sX, sA, (also sU for UDP)
Visualize with IP address vs. the number of
●
packets to unique ports
Copyright (C) 2007 Michael Rash 17
23. Port Sweeps
Packets to a single port number across many
●
hosts
Good evidence of an automated worm (or
●
human) that can exploit a specific vulnerability
in a particular service
Visualize with external IP addresses vs.
●
destination ports vs. packet count
Copyright (C) 2007 Michael Rash 23
28. Slammer Worm
Exploited a vulnerability in MS SQL Server
●
2000
Single 404byte UDP packet to port 1434
●
One of the fastest spreading worms in history
●
Can easily be spoofed – useless to return ICMP
●
port unreachable with your IDS; need to block
the packet with an IPS and patch the
vulnerability
Copyright (C) 2007 Michael Rash 28
31. Nachi Worm
Exploits Windows 2000 and XP systems that
●
are not patched against the MS03026
vulnerability
Always sends a 92byte ICMP echo request to
●
a target system before attempting to exploit the
vulnerability
This ICMP packet makes the worm easy to
●
detect
Copyright (C) 2007 Michael Rash 31
34. Detecting A Compromised System
with Outbound Connections
Honeynet systems should not make outbound
●
connections to external networks (except for
administrative or reporting communication with
known addresses)
SSH and IRC connections are particularly
●
suspicious
Copyright (C) 2007 Michael Rash 34
37. Iptables Logging Args
When building iptables LOG rules:
●
Use logipoptions
–
Use logtcpsequence
–
Use logtcpoptions
–
More attacks can be detected, and operating
–
systems can be passively fingerprinted
Copyright (C) 2007 Michael Rash 37
